Classificaiton models dont run

[knussear]

Trying to run a classification model.

Simple example per your code demo here gbm3::printExamples

data( iris )
kMod <- gbm( Species ~ . , data=iris , distribution = "gaussian", n.tree=1000, shrinkage=.1,
cv.folds=5, train.fraction=.9)
results in
"Error: Not compatible with requested type: [type=character; target=double]."

checked the Species variable, and it is a factor. Not sure how to proceed, seems like an error.

[gregridgeway]

You're asking to gbm to minimize squared error loss (gaussian), but you need a continuous outcome.No gbm3 option for factor outcomesGreg -------- Original message --------From: knussear ***@***.***> Date: 5/1/24 9:40 AM (GMT-05:00) To: gbm-developers/gbm3 ***@***.***> Cc: Subscribed ***@***.***> Subject: [gbm-developers/gbm3] Classificaiton models dont run (Issue #164) Trying to run a classification model. Simple example per your code demo here gbm3::printExamples data( iris ) kMod <- gbm( Species ~ . , data=iris , distribution = "gaussian", n.tree=1000, shrinkage=.1, cv.folds=5, train.fraction=.9) results in "Error: Not compatible with requested type: [type=character; target=double]." checked the Species variable, and it is a factor. Not sure how to proceed, seems like an error. —Reply to this email directly, view it on GitHub, or unsubscribe.You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[knussear]

Yet this is literally directly out of your examples document. Is this no longer possible in gbm? I think the old one used multinomial distributions, but this is no longer an option.

[gregridgeway]

Yikes! Is it really in a gbm3 example? Where is that? I don't see it in the vignettes or the help file.

…

________________________________ From: knussear ***@***.***> Sent: Wednesday, May 1, 2024 10:51 AM To: gbm-developers/gbm3 ***@***.***> Cc: Greg Ridgeway ***@***.***>; Comment ***@***.***> Subject: Re: [gbm-developers/gbm3] Classificaiton models dont run (Issue #164) Yet this is literally directly out of your examples document. Is this no longer possible in gbm? I think the old one used multinomial distributions, but this is no longer an option. — Reply to this email directly, view it on GitHub<#164 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/ACERTQF6HKR6JCMHOE73L6LZAD6PLAVCNFSM6AAAAABHB7R2XOVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDAOBYGU3TSNBTHE>. You are receiving this because you commented.Message ID: ***@***.***>

[knussear]

Yes - Here is where I found it. In the code demos link at the top of the package help there is a link to the print examples, and within that file there is the code I gave above